tonghanwang/ndq

Codes accompanying the paper "Learning Nearly Decomposable Value Functions with Communication Minimization" (ICLR 2020)

ndq's Issues

config file error

Hey,
I try to run the following code:
python3 src / main.py --config = categorical_qmix --env-config = sc2 with env_args.map_name = 2s3z
and
python3 src / main.py --config = tar_qmix --env-config = sc2 with env_args.map_name = 2s3z
I found that due to a config error, it cannot be run directly. I modified the runner: "parallel"-> runner: "parallel_x", and modified the representation of the parameters, such as 1e-2-> 0.01

But in the end it still can't run, can you provide complete and runable config file to help others to reproduce?

Did you encounter this issue?

Training the model on Ubuntu with SC2 4.10.0 and get some replay files. However, when I run

python -m pysc2.bin.play --render --rgb_minimap_size 0 --replay 3m_2020-07-24-05-34-35.SC2Replay

to watch the replay, it failed and returned the following messages.

I0724 18:53:20.919057 173520 remote_controller.py:163] Connecting to: ws://127.0.0.1:23181/sc2api, attempt: 19, running: True
I0724 18:53:23.921703 173520 remote_controller.py:163] Connecting to: ws://127.0.0.1:23181/sc2api, attempt: 20, running: True
I0724 18:53:26.929005 173520 remote_controller.py:163] Connecting to: ws://127.0.0.1:23181/sc2api, attempt: 21, running: True
I0724 18:53:32.214189 173520 sc_process.py:201] Shutdown gracefully.
I0724 18:53:32.214189 173520 sc_process.py:182] Shutdown with return code: 0
Traceback (most recent call last):
  File "C:\Users\me\AppData\Local\Continuum\anaconda3\lib\runpy.py", line 193, in _run_module_as_main
    "__main__", mod_spec)
  File "C:\Users\me\AppData\Local\Continuum\anaconda3\lib\runpy.py", line 85, in _run_code
    exec(code, run_globals)
  File "C:\Users\me\AppData\Local\Continuum\anaconda3\lib\site-packages\pysc2\bin\play.py", line 218, in <module>
    app.run(main)
  File "C:\Users\me\AppData\Roaming\Python\Python36\site-packages\absl\app.py", line 300, in run
    _run_main(main, args)
  File "C:\Users\me\AppData\Roaming\Python\Python36\site-packages\absl\app.py", line 251, in _run_main
    sys.exit(main(argv))
  File "C:\Users\me\AppData\Local\Continuum\anaconda3\lib\site-packages\pysc2\bin\play.py", line 156, in main
    info = controller.replay_info(replay_data)
  File "C:\Users\me\AppData\Local\Continuum\anaconda3\lib\site-packages\pysc2\lib\remote_controller.py", line 71, in _check_error
    return check_error(func(*args, **kwargs), error_enum)
  File "C:\Users\me\AppData\Local\Continuum\anaconda3\lib\site-packages\pysc2\lib\remote_controller.py", line 62, in check_error
    raise RequestError("%s.%s: '%s'" % (enum_name, error_name, details), res)
pysc2.lib.remote_controller.RequestError: SC2APIProtocol.ResponseReplayInfo.Error.ParsingError: 'Could not open initData for the replay: C:\Users\me\AppData\Local\Temp\StarCraft II\TempReplayInfo.SC2Replay'

The version are SMAC==1.0 and PySC2==3.0 and OS==ubuntu, Why using the replay generated by the training process failed? Is there something wrong with the Linux version of PySC2?

It seems that I can run on Ubuntu without rendering but failed on macOS and Windows.

How to create new envs

Hi, I found in your paper, you used new envs which seems to be created by you, how can I create such envs? Are there any tutorials?

Encounter a strange issue

I try to run the code, but I encounter the issue as follows.
Traceback (most recent calls WITHOUT Sacred internals):
File "/usr/lib/python3.5/contextlib.py", line 77, in exit
self.gen.throw(type, value, traceback)
File "src/main.py", line 35, in my_main
run(_run, _config, _log)
File "/home/zhaoyp/pymarl/src/run.py", line 60, in run
run_sequential(args=args, logger=logger)
File "/home/zhaoyp/pymarl/src/run.py", line 227, in run_sequential
learner.train(episode_sample, runner.t_env, episode)
File "/home/zhaoyp/pymarl/src/learners/categorical_q_learner.py", line 184, in train
loss.backward()
File "/usr/local/lib/python3.5/dist-packages/torch/tensor.py", line 198, in backward
torch.autograd.backward(self, gradient, retain_graph, create_graph)
File "/usr/local/lib/python3.5/dist-packages/torch/autograd/init.py", line 100, in backward
allow_unreachable=True) # allow_unreachable flag
RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.cuda.FloatTensor [32, 16, 2, 3]], which is output 0 of SliceBackward, is at version 1; expected version 0 instead. Hint: enable anomaly detection to find the operation that failed to compute its gradient, with torch.autograd.set_detect_anomaly(True).

can not reproduce the results in MMM

Hello, I tried to reproduce your experimental results in MMM. I noticed that you adjusted the built-in AI difficulty to medium(3) in the original paper. I followed this setting but the result is far from the performance in your paper. It's even not as good as the result I achieved under the very hard(7) difficulty. What is your opinion on this?

Unable to reproduce the result of 3b_vs_1h_1m.

I found that bane_vs_hM is indeed the 3b_vs_1h_1m map in the paper, but the result of performance according to the default config is always 0. Can you provide the training hyperparameters for this map?
In addition, do all StarCraft maps use the same default hyperparameters?

How to choose GPU #1 if you execute commands in conda env?

Ask a question:
If you execute the command in docker to select GPU #1:

NV_GPU=1 docker run ......

How to choose GPU #1 if you execute commands in conda env?

python3 src/main.py --config=......